<mripard>
wens: you should probably ask that on #armlinux ;)
<wens>
i guess so :)
leviathancn has joined #linux-sunxi
arokux has joined #linux-sunxi
<wens>
leviathancn: is there supposed to be some firmware running in the background on the a80?
<wens>
beyond arisc, the embedded processor that is.
<wens>
i see references to "smc" calls, but no where is any secure firmware loaded
<mripard>
wens: usually, it's the secure firmware that starts linux
<mripard>
so it's the other way around
<wens>
yeah, but i looked around the bootloader stuff as well
<wens>
no evidence of any secure blob
<leviathancn>
wens: let's say it this way: there are a lot of magic things happening in boot0 ^^
<wens>
that's kind of what i was afraid of....
<Net147>
I get a kernel panic booting mainline inside fdt_check_header. any ideas what it may be?
<leviathancn>
wens: but I have to wait a while to disect the A80 VHDL since the semester in Shenzhen hasn't started and my students visa which allows me to work isn't valid yet
<leviathancn>
^^'
<wens>
is there anyway to tell if i'm in secure mode? besides poking the cp15 register that is...
<wens>
there seems to be a bit of smp code that requires secure mode
<leviathancn>
yes
<wens>
plus the hotplug code pokes secure sram
<mripard>
Net147: it's usually what happens when you don't have a proper DT address given to the kernel
<leviathancn>
when you have two concurrent cores you have access issues
<leviathancn>
they solved it with this risc as the hypervisor
<mripard>
wens: iirc, you don't have access to cp15 in !secure
<leviathancn>
wens,, mripard: exactly
<wens>
mripard: i thought that was configurable, at least for some registers?
<mripard>
which is exactly why you can't really have SMP ops in !secure
<mripard>
which is why PSCI has been developped
<wens>
ok.... but aren't we booting linux non-secure in mainline u-boot? and still have smp on sun6i?
<Net147>
mripard: okay. how do I indicate to the kernel what the DT address is?
<wens>
i probably need to check my configs
<wens>
oh, only boots non-sec on sun7i... no wonder
<mripard>
wens: yeah, we only do so on the A20
<mripard>
the other are booted in secure mode
<mripard>
which makes me think
leviathancn has quit [Read error: Connection reset by peer]
<mripard>
didn't you have some kind of PSCI code for the A31?
<wens>
leviathancn: access issues on? power management?
<wens>
mripard: it hangs when it tries to bring up secondary cores
<wens>
i haven't looked into it more
<wens>
i think HVBAR is fixed to some address, rather than configurable
leviathancn has joined #linux-sunxi
<wens>
leviathancn: access issues on? power management?
<mripard>
HVBAR ?
juanfont has quit [Ping timeout: 245 seconds]
Net147 has quit [Quit: Konversation terminated!]
<wens>
mripard: secure monitor or exception address in hyp mode
<mripard>
ok
<mripard>
and it's fixed to what?
<wens>
0x0
<mripard>
the SRAM boot0 is loaded to ?
gianMOD_ has quit [Remote host closed the connection]
<wens>
would explain why i got u-boot spl output
<wens>
then i tried putting the secure stuff in that sram, and it didn't work
<mripard>
maybe we can just use the same one then?
<mripard>
damn.
<wens>
that's as far as i got
<mripard>
it would be nice to test on the A23 to see if it's the same
<wens>
well it hangs the same, i can tell you that
<wens>
the code is the same, minus the power clamps
<mripard>
weird
<wens>
leviathancn: seems exynos and vexpress all have embedded cores handling power sequencing
gianMOD_ has joined #linux-sunxi
leviathancn has quit [Ping timeout: 272 seconds]
kaspter has quit [Ping timeout: 256 seconds]
naobsd has joined #linux-sunxi
hansg has quit [Quit: Leaving]
dev1990 has joined #linux-sunxi
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
Renard has joined #linux-sunxi
gianMOD_ has quit [Remote host closed the connection]
gianMOD_ has joined #linux-sunxi
gianMOD_ has quit [Remote host closed the connection]
dev1990 has quit [Quit: Konversation terminated!]
Akagi201_ has quit [Remote host closed the connection]
naobsd has quit [Ping timeout: 246 seconds]
nove has joined #linux-sunxi
gianMOD_ has joined #linux-sunxi
ricardocrudo__ has quit [Ping timeout: 272 seconds]
naobsd has joined #linux-sunxi
naobsd has quit [Quit: naobsd]
Akagi201 has joined #linux-sunxi
ricardocrudo__ has joined #linux-sunxi
Black_Horseman has quit [Remote host closed the connection]
a1d3s has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
indy has joined #linux-sunxi
Akagi201 has quit [Ping timeout: 255 seconds]
Andy-D has joined #linux-sunxi
foudubassan has quit [Remote host closed the connection]
foudubassan has joined #linux-sunxi
reinforce has quit [Quit: Leaving.]
pmattern has joined #linux-sunxi
libv has quit [Ping timeout: 272 seconds]
libv has joined #linux-sunxi
ssvb has joined #linux-sunxi
<a1d3s>
hi
<ssvb>
wens: haven't the SRAM and BROM addresses been reshuffled in A80?
<ssvb>
wens: the A80 user manual lists the address map
<wens>
they have
<a1d3s>
ssvb mouse and keyboard works , but no touch , maybe possible that i forget something!?
fredy has quit [Excess Flood]
<a1d3s>
ssvb, your script runs well with 24 depth
<ssvb>
a1d3s: obviously, the kernel needs to have the driver for your touchscreen, and the dts file needs to have it configured
<a1d3s>
driver loaded sun4i-ts
fredy has joined #linux-sunxi
<ssvb>
afaik, this is not the right driver
<a1d3s>
hm this loaded automatic
<a1d3s>
i didnt configure something in /etc/modules
<ssvb>
unless I'm mistaken, it's a driver for the *resistive* touchscreen controller, which is built-in the soc
<ssvb>
and it is never used for real capacitive touchscreens in real devices
<a1d3s>
{ .compatible = "edt,edt-ft5206", }, there adding a line about 5316?
<a1d3s>
im not a programmer ;)
Akagi201 has quit [Ping timeout: 246 seconds]
Akagi201 has joined #linux-sunxi
philippe_fouquet has joined #linux-sunxi
Akagi201_ has quit [Ping timeout: 256 seconds]
philippe_fouquet has quit [Client Quit]
selfbg has joined #linux-sunxi
philippe_fouquet has joined #linux-sunxi
ricardocrudo__ has quit [Quit: Leaving]
ricardocrudo has joined #linux-sunxi
Colani1200|work has quit [Quit: cu]
selfbg has quit [Quit: Leaving]
viccuad has joined #linux-sunxi
leviathancn has joined #linux-sunxi
pmattern is now known as pmattern_afk
JohnDoe7 has joined #linux-sunxi
JohnDoe_71Rus has quit [Ping timeout: 256 seconds]
<leviathancn>
wens: since I don't have the full *A80* spec yet (since I'm still fighting for my study place at 深圳大学) I do not know details about the implementation of their hypervisor firmware
<leviathancn>
otherwise I would already have made an spl for mainline u-boot myself ^^
<wens>
spl is not that hard
<wens>
and the firmware for the embedded processor is loaded by the kernel, evidence in drivers/arisc/
mauro_ has quit [Remote host closed the connection]
khuey has quit [Ping timeout: 264 seconds]
philippe_fouquet has quit []
FR^2 has quit [Quit: Connection reset by peer]
prz has quit [Ping timeout: 264 seconds]
Netlynx has joined #linux-sunxi
Netlynx has joined #linux-sunxi
dev1990 has joined #linux-sunxi
<alexvf>
leviathancn: still no response from allwinner guys? ( about dram documentation and/or shaking screen issue)
<leviathancn>
alexvf: I was too busy translating it. I'm sooo sorry
dev1990 has quit [Ping timeout: 246 seconds]
<leviathancn>
alexvf: I've got an exam the day after tomorrow
<leviathancn>
alexvf: but I'm doing it as part of my todays Chinese study! ^^
<leviathancn>
alexvf: good thing you reminded me
<alexvf>
alexvf: :) no problem at all, just wanted to know if my email got to you
<leviathancn>
alexvf: yeah, got to me
<leviathancn>
^^
<leviathancn>
alexvf: okey, gotta go, lecture is over
<leviathancn>
bye =D
<alexvf>
bye
khuey has joined #linux-sunxi
domidumont has quit [Read error: Connection reset by peer]
<markvandenborre>
I've just read that someone has been able to boot the orange pi plus (allwinner H3 based) using a kernel compiled from public, non-tainted sources
<markvandenborre>
(read it on the mailing list)
<markvandenborre>
anyone knows which devices support boot from nand?
leviathancn has quit [Ping timeout: 256 seconds]
sdschulze has joined #linux-sunxi
p1u3sch1 has quit [Ping timeout: 264 seconds]
p1u3sch1 has joined #linux-sunxi
dev1990 has joined #linux-sunxi
RzR has quit [Excess Flood]
RzR has joined #linux-sunxi
awe00 has quit [Ping timeout: 246 seconds]
diego_r has quit [Quit: Konversation terminated!]
diego_r has joined #linux-sunxi
diego_r has quit [Ping timeout: 245 seconds]
bonbons has joined #linux-sunxi
massi__ has quit [Quit: Leaving]
dantob has joined #linux-sunxi
<hramrach_>
ehlo
<hramrach_>
markvandenborre: technically anything sun4i/sun5i/sun7i based should with some effort
<hramrach_>
depending on what exactly you mean by 'boot from nand'
<hramrach_>
maybe sun6i/sun8i too but I don't follow those
reinforce has quit [Quit: Leaving.]
dantob has quit [Quit: dantob]
ricardocrudo has quit [Ping timeout: 240 seconds]
pmattern_afk is now known as pmattern
<markvandenborre>
hramrach_: the h3 is sun8iw7p1
<hramrach_>
hm, no idea
<hramrach_>
8iw3 and 8iw5 recently grew some quite good support in u-boot but have no idea about 8iw7
<hramrach_>
and nand support is patchy even for the oldest chips
iamfrankenstein has quit [Ping timeout: 240 seconds]
viccuad has quit [Quit: WeeChat 1.1.1]
iamfrankenstein has joined #linux-sunxi
fredy has quit [Excess Flood]
<markvandenborre>
what about sata support?
<markvandenborre>
hramrach_: the h3 has sata...
<markvandenborre>
so a HDD/esata SSD
<markvandenborre>
would also work
<markvandenborre>
I'm trying to avoid shitty performance SSD
<markvandenborre>
both throughput wise and power failure wise
fredy has joined #linux-sunxi
<jemk>
markvandenborre: h3 doesn't have sata
merbanan has quit [Ping timeout: 244 seconds]
dev1990_ has joined #linux-sunxi
Renard has quit [Read error: Connection reset by peer]
Renard has joined #linux-sunxi
ricardocrudo has joined #linux-sunxi
dev1990 has quit [Ping timeout: 256 seconds]
Renard has quit [Read error: Connection reset by peer]
Renard has joined #linux-sunxi
awe00 has joined #linux-sunxi
Netlynx has quit [Quit: Leaving]
domidumont has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
viccuad has joined #linux-sunxi
khuey is now known as khuey|away
dev1990_ has quit [Read error: Connection reset by peer]
dev1990_ has joined #linux-sunxi
jdnurmi has joined #linux-sunxi
pmattern has quit [Quit: Genug für heute.]
<jdnurmi>
Short version: I want my bananapi to have its OTG port run in gadget mode; I've tried a few of the recent branches (sunxi-otg-wip, sunxi-wip) [as of fridayish] and having little to no luck.
<jdnurmi>
The mailing lists seemed to have mixed thoughts around GPIO support for the board (so it was unclear if it should even be possible) -- but if someone could point me at a well-known working gadget-mode build, I'd be more than happy to go away and learn more :)
<jdnurmi>
(tried everything from bananian's 3.4.104->bleeding edge 4.1.. I thought I saw OTG come up in device mode once, but I could never repro except by putting it into FEL mode - so I think I might have accidentally FEL'd or hooked up to my galileo2 by accident (it's OTG works fine))
<jdnurmi>
I'm also super ignorant around FEX's/DTB's, and right now thats where it 'smells' to me that I've got something wrong -- so if anyone wants a guniea pig to guess and check, I'm more than happy to burn out this board if it means the next one will do what I want it to :-D
<jdnurmi>
depending on exactly which dtb/fex/kernel I'm using, I can usually get something like g_serial to insmod, but in those cases I notice "WRN:L2762(drivers/usb/sunxi_usb/udc/sw_udc.c):ERR: usb device is not active" errors (similar if I try to build out a g_multi device); it doesn't show up for enumeration on the host side (except in FEL) - so as far as I can tell, the hardware _should_ be able to do it, it's not some vagueriy I
<jdnurmi>
clearly am underinformed on around allwinner/sunxi/sun7i/bananapi variations that.. to be frank, I could use some of that good old fashioned anonymous internet help around :)
ricardocrudo has quit [Ping timeout: 272 seconds]
jdnurmi_ has joined #linux-sunxi
jdnurmi has quit [Remote host closed the connection]
jdnurmi_ is now known as jdnurmi
khuey|away is now known as khuey
<hramrach_>
jdnurmi: with 3.4/fex setting the port to gadget mode in fex should work
<hramrach_>
in fex *and* kernel config, preferably
<jdnurmi>
where is the best way to obtain the canonical FEX?
<hramrach_>
preferably sunxi-boards
<jdnurmi>
the changelog on bananian mentioned disabling the OTG port (and that's where I started from) -- but it seems that development is still warm on this board, so I'm not sure exactly which repo has the best versions of [kernel, uboot, etc] :)
<hramrach_>
you can also try to extract the fex from your device
<jdnurmi>
now that I've mucked with the device, is that a good idea?
<hramrach_>
the bananapi is so old I have no idea why it still needs a separate repo
<jdnurmi>
hah
<hramrach_>
it's not like the board is special in any way. I could run the same kernel as any sunxi device :s
<jdnurmi>
fair - I'd still like to solve my current problem, but what's the current fan favorite for "can run linux, has a GbE port, and can go gadget-mode"?
<hramrach_>
probably cubietruck
<hramrach_>
but it should work with bpi too
<hramrach_>
if the gadget cable detection is broken or misconfigured it should work when you hardcode gadget mode at least
<jdnurmi>
when I tried to force the module to gadget-only, I got kickback from the kernel saying it was an invalid mode (the switch statement actually doesn't support device-only, at least in the kernel versions I glanced at)
mmarker has quit [Quit: ZNC - 1.6.0 - http://znc.in]
mmarker has joined #linux-sunxi
mmarker has joined #linux-sunxi
<jdnurmi>
since it comes up in FEL, and gets power, I _think_ I've not broken the port, but I'm used to admining x86 boxes, embedded I can run linux on is still a new topic for me :)
<jdnurmi>
lemmie track down that fex and see if I cant make bananian do something OTGish, that'll at least give me a datapoint
<jdnurmi>
(thank you)
<TheLinuxBug>
top -c
<TheLinuxBug>
lol, oops
<hramrach_>
for me the stage/sunxi-3.4 branch kernel gives the option to switch to device only
<hramrach_>
but I'm not running it on anything a20 atm
<jdnurmi>
I get the option, but if I run it, dmesg kicks back
ricardocrudo has joined #linux-sunxi
<jdnurmi>
(I'll dig out the code in a moment, but..) so using the fex from https://github.com/linux-sunxi/sunxi-boards/blob/master/sys_config/a20/Bananapi.fex - the OTG port looks like it should already be active in that config. I do get the cpu/load-avg issue when detect = 1, can insmod, but still get "usb device is not active" from it
<jdnurmi>
ah "Invalid or missing 'dr_mode' property" - if you search for that - at least the WIP I was testing the other day doesn't actually work if you configure for device-only mode
<jdnurmi>
(I think that was the sunxi-otg-wip branch, but wouldn't swear to it)
<hramrach_>
hmm, I have 3.4.75 kernel on some board and it's broken
<jdnurmi>
I'm pretty good at abusing linux in weird ways, but my embedded knowledge is... less strong than I need it to be -- if I can be your hands on a board/kernel/test, I'm happy to do that :) (while I go and find the apropriate cubieboard)
<jdnurmi>
over all the versions and iterations I've tested, there seems to be some disagreement around the right "PH#" ports for the vbus -- but other than noticing the discrepency, I've never even been sure if they're meaningful
<jdnurmi>
(nor really what it means other than that I need to learn more about GPIO)
<jdnurmi>
(specifically around PH05/PH15 and some third GPIO option)
TheLinuxBug has quit [Ping timeout: 246 seconds]
Turl has quit [Ping timeout: 246 seconds]
TheLinuxBug has joined #linux-sunxi
<hramrach_>
you probably don't need vbus in device mode
Turl has joined #linux-sunxi
<NiteHawk>
shouldn't that read: you don't want (to drive) vbus in device mode?
<jdnurmi>
while not trying to instigate a flamewar, cubietruck doesn't seem to support GbE - my hope is to pass somewhere 100Mbps < n < 300 Mbps via GbE -> g_mass_storage -- the galileo can get pretty close to the 100Mbps, but my want is a bit closer to 150 - with a CPU / OTG port that can handle the other half as well
<hramrach_>
jdnurmi: the trick is it works with built-in gadget but not module
<hramrach_>
so if you build your gadget into the kernel it should work on bpi as much as ct
libv has quit [Ping timeout: 264 seconds]
bsdfox has quit [Ping timeout: 240 seconds]
libv has joined #linux-sunxi
libv has quit [Ping timeout: 272 seconds]
libv has joined #linux-sunxi
<jdnurmi>
lemmie give that a spin.. harder to test w/o module arguments, but let's break something!
bsdfox has joined #linux-sunxi
<jdnurmi>
hmm, ok, so this is going to be a source-build w/ a custom FEX - I've not done that combo yet; for trying to match anything you're checking, what kernel build are you using?
<jdnurmi>
(and tangentially, anyone know a good way to preserve an configfs device? tar screams unhappiness and shell works for me, but doesn't seem like quite the right answer either)
viccuad has quit [Quit: WeeChat 1.1.1]
libv has quit [Ping timeout: 272 seconds]
<hramrach_>
jdnurmi: I use the ethernet gadget which has no arguments
<hramrach_>
just ifconfig it on both ends
<hramrach_>
the gadgets that require configuration mostly fail
<hramrach_>
there is probably some mismatch in the different kernel pieces merged together
<hramrach_>
and nobody cares enough to debug and fix everything when some simple gadgets still work
<jdnurmi>
probably unrelated, but I found I coudln't build the allwinner USB parts as modules themselves (some unexported symbol that existed when compiled in)
<hramrach_>
yes, that's known issue
<hramrach_>
this will probably work on mainline
<hramrach_>
once the driver is ready
nove has quit [Quit: nove]
libv has joined #linux-sunxi
<hramrach_>
for mass storage there is some way to transform module arguments in kernel commandline arguments so it should still be doable
<jdnurmi>
for now let me rebuild w/ ether - if one works it at least proves the pipeline and eliminates a large unknown for me :)
<hramrach_>
g_ether gives most flexibility but g_serial with ppp/slip might be simpler and faster. Never tried it
<hramrach_>
and you like don't need storage gadget when you have sshfs
<jdnurmi>
well, really I want the g_storage gadget to feed a random vendor device that wants a USB key -- so I'm feeding it a dm-cached volume fed from my SAN -- but it's getting a working g_mass_storage gadget against a gigabit pipe that I'm having a hard time finding the hardware/software combo to quite bang it out
<hramrach_>
hmm, then you really need the storage part
<jdnurmi>
yea, the galileo2 did it /technically/ but just didn't have the CPU to handle the acutal bandwidth
<hramrach_>
there is some usb support in the sunxi-wip branch iirc but no idea how that works. never really tested that
<hramrach_>
and you need to convert your fex to DT
<jdnurmi>
I do weird things, so clearly, don't optimize for my needs - but I'm happy to test weird things nobody else ever seems to want :)
<hramrach_>
so better get something that works with 3.4 to start with I guess
frank has quit [Ping timeout: 264 seconds]
<jdnurmi>
I've got a different sdcard w/ a dtb bootup, but was equally confused as to who's dtb was really authoritative fot the bPI
<jdnurmi>
figured it was easier to ask for support on the FEX since it at least had host-side USB for the OTG alleged
<jdnurmi>
depending on how I pivoted the vusb/detect values, I've gotten a 4.x+ kernel + dtb to boot, but since I barely understand what I'm doing in the dtb end, I've been hesitant to do much more than quiltwork on patches I found
<jdnurmi>
I'm probably reaching, but is there an etch-a-sketch mode for the bPI -- return to factory/vendor state? If everyone else swears it works, then surely I'm doing it wrong?